Automated reconstruction of whole-embryo cell lineages by learning from sparse annotations
Caroline Malin-Mayor1, Peter Hirsch2,3, Leo Guignard1,4
1HHMI Janelia, Ashburn, VA, USA.
Nature Biotechnology
|September 6, 2022
Summary
We developed a new method to automatically track cell nuclei in developing embryos using deep learning. This approach significantly improves the reconstruction of cell lineages, aiding the study of cell fate decisions.
Area of Science:
- Developmental Biology
- Cell Biology
- Bioinformatics
Background:
- Accurate tracking of cell lineages is crucial for understanding embryonic development.
- Existing methods struggle to reconstruct long cell lineages in time-lapse microscopy data.
- Identifying cell fate decisions requires precise monitoring of individual cells over time.
Purpose of the Study:
- To present a novel computational method for automated nuclei identification and tracking in whole embryo microscopy recordings.
- To improve the accuracy and completeness of cell lineage reconstruction.
- To enhance the understanding of spatiotemporal cell fate decisions during development.
Main Methods:
- Combining deep learning algorithms with global optimization techniques.
- Applying the method to time-lapse microscopy recordings of entire developing embryos.
- Utilizing a mouse dataset for validation and comparison.
Main Results:
- Successfully reconstructed 75.8% of cell lineages over a 1-hour period in mouse embryos.
- Demonstrated a significant improvement compared to competing methods, which reconstructed only 31.8% of lineages.
- Validated the method's efficacy in handling complex, dynamic biological systems.
Conclusions:
- The developed method offers a robust solution for automated cell lineage tracing in developmental studies.
- Improved lineage reconstruction facilitates a deeper understanding of cell fate determination in embryos, tissues, and organs.
- This technology has the potential to advance research in developmental biology and regenerative medicine.
